Parents have been lied to for twenty years. Every social network, every game, every device promised it was "safe" and most of them weren't. By the time AI showed up, the parent on the other side of the marketing page had her defences up to her eyebrows. If your messaging sounds like the same script she has read a hundred times, you lose. If it sounds like a legal disclaimer, you also lose.
This is how we coach kid-safe AI teams to talk about their product without either trying to charm parents into trusting them or burying every claim under so many caveats no one knows what the thing does.
Principle 1: Lead with what the product won't do
Adult AI marketing leads with capability - what it can write, generate, solve. Kid-safe AI marketing should lead with restriction - what it won't say, what it won't ask, what it won't remember. Parents are scanning for boundaries before features. Give them the boundaries first; they will read the features second.
Principle 2: Name the safety layer, don't hide behind the model
No foundation model is safe for children out of the box. Every team in this space knows this. Pretending otherwise - saying "powered by GPT, safe by design" - makes you look either naive or dishonest. Instead, describe the system you built around the model: the filters, the policies, the age-awareness, the human review, the parental controls. That system is the product. The model is a component.
Principle 3: Use specific, falsifiable claims
"Safe for kids" is not a claim, it's a vibe. "Blocks 99.4% of self-harm prompts across 12 languages, independently benchmarked" is a claim. Specifics terrify marketing teams because they can be checked, and that is exactly why parents trust them. Pick a small number of measurable claims you are willing to stand behind, publish the methodology, and let people verify them.
Principle 4: Be honest about the failure modes
You will not catch everything. Say so. Say what happens when you don't catch something - who reviews it, how fast you respond, how parents are told. The first time a journalist writes about an incident at your product, the difference between a story and a non-story is whether you had already told the world this could happen and how you handle it when it does.
Principle 5: Show, don't claim
Wherever possible, let parents see the product behave. A live demo where they can try to break it. A public archive of the prompts your safety layer blocks. A redacted incident log. Screenshots of the parent dashboard. Words are cheap; product is expensive - leading with product builds trust no copy ever can.
What not to say
- "Completely safe." Nothing involving generative AI is completely safe. Parents know.
- "Powered by [frontier model], so it's safe." The model is not the safety story.
- "COPPA-compliant" used as a marketing claim. It's a legal floor, not a value prop.
- "Designed with kids in mind." Empty phrase. Show, don't tell.
- "AI for everyone, including kids." If it's also for adults, it's not really for kids.
What to say instead
- What the product will not do, in plain language.
- Who reviewed the safety system and what they found.
- What you measure, how often, and where parents can see it.
- How a parent regains control when something goes wrong.
- Which independent benchmark or certification you've published against.
"Parents don't want marketing. They want a product that behaves like the brochure said it would. Write a brochure you can actually live up to."